FP context switch xtest - #828
Open
dave-patel-riscstar wants to merge 3 commits into
Open
Conversation
Introduce riscv_fp_ctx.h, a small assembly-based helper library used by both xtest and the os_test TA to observe the behaviour of floating‑point registers across a call. The goal is to verify correct FP context preservation without interference from compiler optimisations. Only callee‑saved FP registers (fs0..fs11) and fcsr are covered. Caller‑saved register (ft0..ft11, fa0..fa7) are intentionally excluded since the ABI allows a callee to clobber them. The routines are implemented in assembly to ensure value remain in FP registers rather than being spilled to the stack by the compiler. They are emitted from a header instead of a .S file so both xtest and the TA can share the same code without adding assembler rules to their respective build systems. Signed-off-by: Dave Patel <dave.patel@riscstar.com>
Add TA_OS_TEST_CMD_RISCV_FP_CONTEXT, which runs a set of sub-tests that
each install a known floating-point context, make a different kind of
excursion out of the TA, and check that the context comes back intact.
A TA runs with the floating-point unit disabled and is handed a context on
the first floating-point instruction it executes, so every sub-test starts
by taking that trap. What they then cover is:
SYSCALL a syscall serviced entirely inside the TEE, where the TA
context is saved on the way in and has to be given back on
the first floating-point instruction after the return
RPC TEE_Wait(), which suspends the thread, runs the normal
world and resumes through thread_resume_from_rpc()
CRYPTO a digest, which the TEE may compute in a secure kernel
floating-point section of its own and so has to take the
registers from the TA and hand them back
TAINT leaves a pattern in the registers and returns
CHECK_TAINT reads the registers before writing any of them and fails
if an earlier TA's pattern is still there
On failure the index of the first field that did not survive is reported
back to the caller so that a failure says which register was lost.
Signed-off-by: Dave Patel <dave.patel@riscstar.com>
Drive the os_test sub-tests from xtest and add the checks that can only be made from the normal world. Alongside the TA side cases the test holds a pattern in this process' fs0..fs11 and fcsr across a call into the TEE. That is the normal world half of the domain switch, and it is run twice: once against a TA command that uses floating point and once against one that does not, since the TEE restores the normal world context in the first case and skips the restore in the second, having never disturbed the registers. Both have to leave the caller's registers exactly as it left them. A concurrency case runs the RPC sub-test from several threads at once, since the floating-point bookkeeping is per OP-TEE thread and one thread's context must not surface in another. The test skips itself when the TA reports TEE_ERROR_NOT_SUPPORTED, which covers both a non-RISC-V target and an OP-TEE built without CFG_WITH_VFP. Signed-off-by: Dave Patel <dave.patel@riscstar.com>
dave-patel-riscstar
force-pushed
the
rp022-m7-fp
branch
from
September 1, 2026 18:22
92fd17a to
8c09317
Compare
Author
|
Add TA_OS_TEST_CMD_RISCV_FP_CONTEXT, which runs a set of sub-tests that A TA runs with the floating-point unit disabled and is handed a context on SYSCALL a syscall serviced entirely inside the TEE, where the TA |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
xtest for FP
A TA runs with the floating-point unit disabled and is handed a context on
the first floating-point instruction it executes, so every sub-test starts
by taking that trap. What they then cover is: